[Paraview] ParaView 1.4.3 in render-server mode

[Joseph Insley]

Berk and Amy, 

Thanks for your help, things seem to be running just fine now.  Although, 
I would like make sure I understand how things work in this mode.  If I 
understand correctly, when you run with just a client and a server, the 
server computes the geometry and ships it to the client, who does all of 
the rendering.  With the client, server, render-server model the server 
computes the geometry and ships to the render-server, who does the 
rendering, then ships pixels to the client for display.  Is that about 
right?

Thanks again for your help,
joe.

On Thu, 2 Sep 2004, Berk Geveci wrote:

>
>Hi Joe,
>
>> Without more detailed documentation I don't fully understand what is 
>> happening, but it looks to me that the data server is trying to connect to 
>> the localhost, rather than myRenderNode.  Do I need to give the data 
>> server this information, or should it be getting it from the client?  I 
>> tried playing around with using the --machines option, but didn't 
>> see any difference, then again I wasn't sure of what the syntax of the 
>> file should be.
>
>You are on the right track. You have to use the --machines=filename
>option. However, you should put this option on the render server command
>line not the client or data server. The file should contain the
>hostnames of all render server nodes (newline separated).
>
>> Also, when I tried running with multiple nodes for both servers, the first 
>> render node listens on the port I specified in the --render-node-port 
>> option, but the rest use ports presumedly generated by the render server.  
>> When I tried passing the --render-node-port a comma separated list of 
>> ports, only the first one was used.
>
>As far as I know, there is no way of specifying which ports to use on
>the render server nodes. ParaView will ask libc to come up with
>available ports and use them. As long as you set the machines file
>properly, this shouldn't be a problem (unless you have a firewall that
>blocks those ports)
>
>-Berk
>
>

====================================================================
joseph a. insley				insley at mcs.anl.gov
mathematics & computer science division		(630) 252-5649
argonne national laboratory                     (630) 252-5986 (fax)
http://www.mcs.anl.gov/~insley
====================================================================